Transcriptome Dynamics From Primary Gastric Cancer to Peritoneal Metastases: Translational Insights for Surgical Oncology.
BACKGROUND AND AIMS Peritoneal metastases (PMs) represent the most frequent, clinically challenging dissemination pattern in advanced gastric cancer, especially poorly cohesive (PCGC) subtype, and are associated with poor prognosis. We aimed to characterize the transcriptomic landscape of primary gastric tumors (PGTs) and matched PMs to identify differential molecular programs, tumor-microenvironment (TME) features, and potential biomarkers. METHODS We analyzed 55 FFPE samples from 23 treatment-naïve patients with synchronous peritoneal-only stage IV PCGC and 10 non-neoplastic gastric controls (NNC). RNA sequencing was analyzed using DESeq2 and GSEA, with cell proportions estimated through immune deconvolution (TIMER/xCell). RESULTS A total of 4279 differentially expressed genes were identified across all groups. CLDN18 expression progressively decreased from NNCs to PGTs and PMs. Direct PGT-PM comparison revealed that PGTs were characterized by cytoskeletal and extracellular-matrix remodeling genes upregulation, mitotic/cell-cycle, NOTCH and apical-junction pathways enrichment, and memory T-cell predominance. Compared with PGTs, PMs showed increased IGF1, IGFN1, NTRK2 and adipogenesis-related transcripts, adipogenesis and MAPK7/11-NTRK2 signaling enrichment, and macrophage-dominant and dendritic-cell-dominant microenvironments. epithelial-mesenchymal transition (EMT), inflammatory pathways, and cancer-associated fibroblasts were shared by both tumor lesions. PMs did not match established TCGA or PCGC subtypes, suggesting a distinct molecular identity. CONCLUSIONS PMs are characterized by TME remodeling and transcriptional reprogramming. Shared EMT and inflammatory pathways act as bridge linking PGT to PM, consistent with adaptive exploitation of physiological peritoneal programs. Collectively, our findings suggest a transcriptomic framework in which PM can be characterized as a distinct niche-conditioned biological entity and support microenvironment-informed diagnostic and therapeutic strategies.